An automatic detection method for remote sensing moored ships based on wharf spectral information

Abstract

The invention discloses a remote-sensing automatic detection method for berthed ships based on the spectrum information of a harbor, and the method comprises the steps: carrying out the sea-land segmentation through a spectral band ratio value; carrying out the harbor extraction along a sea-land boundary line based on common spectrum features of harbors; detecting suspected ships or ship groups nearby the harbor; carrying out the histogram projection of the suspected ships or ship groups in a vertical direction of the harbor; segmenting a plurality of ships or ship groups which are berthed ina longitudinal column; finally carrying out the histogram projection in a parallel direction of the harbor, and segmenting the ships which are berthed in parallel. The method effectively solves a problem that the harbor and ships are difficult to discriminate through a conventional remote sensing image, and can achieve the full-automatic detection of stable berthed ships.

Description

technical field [0001] The invention relates to a remote sensing automatic detection method for moored ships based on wharf spectral information, and belongs to the application field of remote sensing image intelligent recognition. Background technique [0002] The main difficulties in the automatic detection of berthing ships are: 1) The difference in image features between the ship and the sea water is relatively small, and it is difficult to obtain satisfactory results through automatic processing; 2) The image features between the wharf and the ship are similar, so it is difficult Automatic distinction; 3) Multiple ships can be parked at the same dock, and it is difficult to automatically distinguish between ships. [0003] There are two main traditional automatic detection methods for parked ships: one is to rely on the image features of the ship, and to detect the ship through a full-image search, which is difficult to obtain stable detection results due to image grays...

AI Technical Summary

Problems solved by technology

[0002] The main difficulties in the automatic detection of berthing ships are: 1) The difference in image features between the ship and the sea water is relatively small, and it is difficult to obtain satisfactory results through automatic processing; 2) The image features between the wharf and the ship are similar, so it is difficult Automatic distinction; 3) Multiple ships can be parked at the same dock, and it is difficult to automatically distinguish between ships

[0003] There are two main traditional automatic detection methods for parked ships: one is to rely on the image features of the ship, and to detect the ship through a full-image search, which is difficult to obtain stable detection results due to image grayscale changes, shadow changes, and ship structure changes; It relies on the wharf image template, and performs change detection to extract ships through the wharf template registration, but the detection results cannot be obtained in unknown berthing areas

Images